Manual Meeting Follow-Ups Cost 134 Hours and €4,700 Per Year: The Digiotouch AI Pipeline That Automates It With Dynamic Workflows

Manual meeting follow-up costs a business roughly 134 hours and €4,700 a year per employee, combining the 2.8 hours a week Asana attributes to unnecessary meetings with the €34.90 average hourly labour cost Eurostat records for the EU. The answer is not more discipline but a repeatable pipeline: capture the meeting once, work from a structured record of decisions and action items, then let a dynamic workflow deliver the output to the tools your team already uses. Digiotouch AI records online meetings from the desktop app and in-person meetings from the phone, returns a summary and action items on every plan, and adds the full transcript from the Essential plan. What changes is the job itself, because you stop writing from a blank page and start reviewing a draft that already exists.

Claire runs an independent consultancy and takes four client calls a day. She listens properly during the call, which means she writes almost nothing down. By Friday she has many recaps to send, and she is reconstructing Tuesday afternoon from three lines of notes and whatever she can still remember. Two of the four go out late. One of them gets a decision slightly wrong, and the correction costs her another half hour and a small amount of credibility.

The expensive part of a meeting happens after it ends

Most advice about meetings is about the meeting. Shorten it, invite fewer people, set an agenda. All of that is sound, and none of it touches the part that actually eats the week, which is what happens once the call is over and someone has to turn an hour of conversation into something another person can act on.

The scale of the underlying problem is well documented. In Stop the Meeting Madness, Leslie A. Perlow, Constance Noonan Hadley and Eunice Eun report that executives spend an average of nearly 23 hours a week in meetings, up from under 10 hours in the 1960s. Of the 182 senior managers they surveyed, 71 percent called meetings unproductive and inefficient, and 65 percent said meetings kept them from completing their own work.

The follow-up sits on top of that. Asana's Anatomy of Work Global Index 2023, a survey of 9,615 knowledge workers conducted by GWI, found that coordination rather than skilled work takes up 58 percent of the working day, and that unnecessary meetings alone drain 2.8 hours a week from knowledge workers and 3.6 hours from those in senior leadership roles. Respondents estimated that better processes would give them back 4.9 hours a week.

Put a price on those hours and the problem stops being an annoyance. Eurostat put average hourly labour costs in 2025 at €34.90 across the EU and €38.20 in the euro area. At 2.8 hours a week over 48 working weeks, that is 134 hours and about €4,700 a year, for one person. The body below works that calculation out in full, and shows you how to run it on your own rate.

Where does the time after a meeting actually go?

We asked. In a Digiotouch AI survey of 43 professionals, fielded internationally in November 2024, 76.8 percent rated meeting notes 4 or 5 out of 5 for usefulness. The open-text answers told the other half of the story. Respondents described writing those notes as tedious, easy to lose focus during, and often badly structured by the time they were finished. One person called it their worst task.

That gap is the whole problem in one line. The output is wanted. The labour is not. And because the labour is the part a person can skip without anyone noticing until later, it slides to Friday, then to next week, then into memory, where the detail quietly degrades.

The same survey found that 69.8 percent of respondents would use a secure AI tool that writes meeting notes in seconds, while about 5 percent would not. With a sample of 43, those figures show direction rather than precise population values, but the direction is not subtle. More telling is what people asked for. The top requests were action items assigned to specific team members, and note formats that fit the type of meeting. Almost nobody asked for a better transcript on its own. What they wanted was follow-up they could act on.

That is the difference between a recording and a pipeline. A recording hands the raw material back to you. A pipeline hands you the finished thing.

How Digiotouch AI handles this: recording runs in the background while you take part in the conversation normally. Voice, video and the slides shown on screen are all captured, so the summary reflects the figure somebody put up rather than only the sentence they said about it. Nobody has to type during the meeting for the record to exist afterwards.

How do you capture a meeting so the follow-up writes itself?

The capture step decides the quality of everything downstream, and the right method depends on the kind of meeting rather than the device you happen to prefer.

For an online meeting, the desktop app records what happens on your machine. Nothing is added to the participant list on desktop, so the attendee list your client sees is the one they expect. Sessions run up to two hours, with a warning at one hour fifty-five, which covers a workshop but not an all-day offsite.

For an in-person meeting, the mobile app records the room through the phone microphone. Put the phone flat on the table between the participants rather than beside one person, because speaker attribution depends on clean audio.

For a file you already have, import it. Digiotouch AI accepts mp4, webm, mp3, wav, mpeg and m4a, and treats an import the same way as a live recording. Imports draw on both your storage and your monthly minute allowance, so they are not free in the way an unlimited label might suggest.

One point of etiquette that is also a legal obligation in most places: telling participants you are recording is your job, not the software's. Digiotouch AI keeps a participant consent reminder visible for the duration of the recording, and recording does not begin on its own. Automatic start at the time of a calendar meeting exists, but it is opt-in and off by default.

How Digiotouch AI handles this: the desktop app captures online meetings without joining the call as a participant, the mobile app captures the room in person, and the web app gives you the full library in a browser with nothing installed. It is one account and one library across all three, so where you recorded stops mattering the moment the meeting ends.

What should a structured meeting record contain?

A usable record answers four questions without anyone having to listen back: what was discussed, what was decided, who owes what by when, and what is still open.

Most notes handle the first and fumble the rest. Decisions get written down as topics, so six weeks later the file says the team discussed pricing and not what they settled on. Action items get written down without an owner, which means they belong to everyone and therefore to nobody. Open questions vanish entirely, because people rarely record the thing that was left unresolved.

Attribution is the piece that turns a summary into something you can hold up in front of a client. Knowing that a scope change was proposed by the client rather than volunteered by you is not a nicety when the invoice is queried. Speaker identification works differently by platform, and it is worth knowing which one you are relying on. On desktop it combines audio with video window activity. On mobile it leans on participants introducing themselves out loud at the start, which takes ten seconds and repays itself.

Two honest limits. Simultaneous speech degrades attribution, as it degrades any transcription. And video is processed for on-screen material only: there is no facial recognition, no identification of individuals from video, and no analysis of emotion, sentiment or tone of voice.

How Digiotouch AI handles this: every plan returns a structured summary with action items, including assignees and deadlines drawn from what was said in the conversation and marked TBD where nothing was stated. Key topics and the full transcript come with Essential, and decisions and open items are added on Professional. Every field stays editable, so correcting one owner does not mean rewriting the document.

How do you turn a meeting record into a client-ready deliverable?

A structured record is not the deliverable. It is the raw material the deliverable is built from, and this is the step where most of the remaining time goes.

In practice there are three outputs a consultant needs from most client calls. A recap email confirming what was agreed, sent the same day while it still carries weight. A set of action items assigned to named people on both sides. And, after a discovery call, the raw input to a proposal, which is mostly the client describing their problem in their own words, a thing you want quoted accurately rather than paraphrased from memory.

The method is the same for all three. Start from the summary rather than the transcript, because the summary is already organised. Ask the Digi assistant for what you need from the meeting, for example the client's stated constraints, or every point where a deadline was mentioned. Then edit. The draft carries the facts and you supply the judgement, which is the correct division of labour and the reverse of how most people currently work.

The Digi assistant answers from everything that was captured, including the slides shown on screen, so a question about the figure on the third slide is answerable even though nobody read that figure aloud. Where you need exact wording from the conversation, that is where the transcript earns its place, and the transcript is an Essential feature.

How Digiotouch AI handles this: the summary, action items and transcript sit together in one meeting record, and the Digi assistant answers questions across all of it. Output goes out as PDF or DOCX on any plan, or by a share link. One limit worth knowing before you rely on it: a shared meeting opens under Shared with me in the recipient's own Digiotouch AI account, so it is not a public link you can send to anyone.

Where do the action items go afterwards?

A recap that lives inside a meeting tool is a recap nobody will look at. The last step of the pipeline is delivery, and this is what dynamic workflows do.

A workflow is a route you configure once. It takes the summary, decisions and action items from a meeting and delivers them to the tools your team already works in, such as Asana, Notion, Slack, Trello, Confluence or SharePoint. Set it up before the meeting rather than after, because a workflow configured on Friday does not retrospectively deliver Tuesday.

The allowance scales with the plan. Free includes one workflow with a single export destination, which is enough to prove the idea works with one tool. Essential raises that to five workflows across up to ten destinations. Professional runs fifteen workflows with unlimited destinations, which is the point at which different meeting types can genuinely route to different places. Current allowances are listed on the pricing page.

What this changes in practice is small and significant. The action item does not need copying out. It arrives where the work is tracked, with the owner and the deadline already attached, and the follow-up stops depending on somebody remembering to do the copying.

How Digiotouch AI handles this: dynamic workflows are available from the free plan, with the number of workflows and destinations rising through Essential and Professional. Export to PDF and DOCX is included on every plan, so the record is portable even before a workflow is configured.

What does manual follow-up cost you, and how do you work it out?

Here is the arithmetic in full, so you can check it rather than take it on trust.

Take the hours a week lost to meeting overhead from Asana's index, multiply by the Eurostat hourly labour cost, then multiply again by 48 working weeks. Nothing else goes into it.

Sources: hours from Asana's Anatomy of Work Global Index 2023, hourly labour costs from Eurostat for 2025, both linked in the citation index. The 48-week year is our assumption and allows for leave and public holidays. At Asana's own 52 weeks, the knowledge worker row becomes 146 hours and roughly €5,100.

To run it on yourself, use your own billing rate rather than a labour cost average, since for a consultant the relevant number is what the hour could have earned instead. Count the meetings you write up in a normal week, estimate the minutes each one takes including the ones you redo, and multiply. Most people are surprised by the total, because no single instance feels expensive. It is a rounding error, forty-eight times a year.

How Digiotouch AI handles this: we are not going to attach a saving to that figure. We have not measured what our own users recover, and a number we cannot show you the working for is worth nothing. What we can say is what changes structurally: the write-up becomes a review, and reviewing a draft that already holds the decisions and action items takes less time than producing one from memory. Measure it on your own week and the number will actually be yours.

What still needs your own review?

A pipeline nobody checks is a liability, and an article that pretends otherwise is selling something. Four things still need you.

The summary, before it reaches a client. Generated text is accurate about what was said and is not a judge of what mattered. A point a client made in passing may be the most important thing in the call, and a long tangent may deserve one line. That weighting is yours.

Action item owners. Deadlines and assignees come from what was actually said in the meeting, so anything left unstated is marked TBD rather than guessed at. TBD is honest, and it is also a queue of decisions waiting for you.

Specialist vocabulary. Product names, acronyms particular to one client and unusual proper nouns are where any speech recognition is weakest. Skim for them.

And the recording itself, occasionally. On mobile, if the phone battery dies mid-recording, the recording ends and cannot be recovered. On desktop, the local copy is deleted once the recording and the summary have uploaded successfully, so the account is the record rather than your hard drive. Uploads are journalled, which means a failed upload retries at next launch instead of losing the meeting.

How Digiotouch AI handles this: every field in the summary and the action item table is editable, and edits survive a re-export. The pipeline assumes a human reviews the output before it leaves the building, which is why nothing goes anywhere until you configure a destination.

What does a weak follow-up cost you six weeks later?

The hours are the visible cost. Retrieval is the expensive one.

Atlassian's State of Teams 2025, a survey of 12,000 knowledge workers and 200 executives, found that teams waste around a quarter of their time simply searching for answers. Not producing them. Finding ones that already exist somewhere.

That is what a weak follow-up buys you later. A client asks whether the integration was in scope. You know it was discussed. You do not know what was concluded, and the recap you sent in a hurry on a Friday says the team reviewed scope. So you either reconstruct it, which is guessing with extra steps, or you concede the point, which has a price attached.

A structured record avoids this because it is searchable text rather than an audio file nobody will ever open. Decisions are stored as decisions. Six weeks later the question takes seconds rather than an afternoon, and it is answered from what was said rather than from what everyone now remembers being said. This is also the argument for routing meeting output into the tools where your team already looks for things, rather than leaving it in a meeting library, because it then becomes a searchable base of what your team has actually decided.

How Digiotouch AI handles this: the meeting library is searchable across summaries, action items and transcripts, and the Digi assistant answers questions across past meetings, including what was shown on screen. Meeting content is never used to train AI models.

Manual follow-up compared with a configured pipeline

The same four client calls a week, handled two ways. This compares the job rather than the tools, and it assumes in both cases that a person reviews the output before it goes out.

Sources: Digiotouch AI product behaviour as described above, and plan allowances as published on digiotouch.ai/en/pricing.

Key takeaways

• The cost is real and calculable. At 2.8 hours a week and €34.90 an hour, meeting follow-up runs to about 134 hours and €4,700 a year per person, from Asana and Eurostat figures you can check yourself.
• People want the record, not the writing. In our survey of 43 professionals, 76.8 percent rated meeting notes highly useful while describing the writing of them as a chore, and 69.8 percent would hand the job to a secure AI tool.
• Capture method follows meeting type. Desktop for online meetings, mobile for in-person, web for working in a browser, all on one account and one library.
• Delivery is the step people skip. A dynamic workflow routes the summary and action items to the tools your team already uses, from the free plan upwards, and it has to be configured before the meeting rather than after.
• Review is not optional. Weighting what mattered, resolving TBD owners and checking specialist terms remain yours, and any tool claiming otherwise is overselling.

Next step

Test the pipeline on one recurring meeting rather than all of them. Record it, review the summary, configure a single destination, and see what your own week looks like after four of them.

Related reading

• 7 in 10 professionals would let AI take their meeting notes: the survey behind the opening section of this article, including the full feature wish list.
• How to turn a recorded meeting into a shareable summary: the mechanics of the deliverable step, from three different starting points.
• Turn meetings into a searchable knowledge base automatically: what happens to the record after delivery, and why retrieval is the argument beyond hours saved.

Sources

• Eurostat, EU hourly labour costs ranged from €12 to €57 in 2025, 31 March 2026.
• Leslie A. Perlow, Constance Noonan Hadley and Eunice Eun, Stop the Meeting Madness, Harvard Business Review, July to August 2017.
• Asana, Anatomy of Work Global Index 2023, 8 March 2023.
• Atlassian, State of Teams 2025.
• Digiotouch AI, 7 in 10 professionals would let AI take their meeting notes, 17 August 2026.